Source: Dissertation Abstracts International, Volume: 79-12(E), Section: B.

Adviser: Albert I. Ko.

Access restricted by licensing agreement.

This item is not available from ProQuest Dissertations & Theses.

Leptospirosis is a major zoonotic source of morbidity and mortality worldwide. While it was traditionally considered a disease of rural farmers and remains a problem in that setting, it has also recently emerged as a public health issue in urban slums. Despite the high burden of leptospirosis around the world, important questions remain about both regional and individual-level transmission dynamics and infection determinants. This dissertation combines epidemiology and advanced quantitative methods to answer key questions about leptospirosis across spatial and temporal scales.

My first chapter characterized the spatio-temporal determinants of leptospirosis during an epidemiologic transition in Thailand. Thailand experienced an explosive countrywide emergence of leptospirosis in the late 1990s that established endemic transmission in the agricultural northeastern region of the country. This persistent transmission has proven difficult to manage due to a lack of understanding of the spatial and temporal determinants of infection. I analyzed the relationship between 15 years of weekly reported cases and weather, landscape, and agricultural features at the district level. My goal was to provide information at policy-relevant spatial and temporal scales that could be used to target interventions. I identified that. in contrast to the individual level, rural leptospirosis at the district level is not strongly associated with rice farming. Neither temporal nor spatial analysis identified rice cultivation as a major driver of reported cases. In the absence of rice, I identified rainfall and temperature in recent weeks, as well as other characteristics of the landscape and agricultural system, as determinants of the spatially heterogeneous incidence. District-level infection determinants varied between the emergence and subsequent endemic period.

My next two chapters focused on leptospirosis in urban slum residents. Through detailed longitudinal studies in Pau da Lima, an urban slum community in Brazil, our group has identified a number of household risk factors for infection. However, these are a static measure of risk, which is actually a dynamic function of interaction with the contaminated environment. I used GPS tracking to quantify the movement patterns of urban slum residents and their resulting exposure to leptospirosis transmission sources. I identified that urban slum residents spend the majority of their time within 50 meters of the home. Additionally, residents across age and gender groups had the same amount of movement-induced exposure to transmission sources in the environment. I did, however, find significant differences in the activity space---area visited per 24-hour period---between risk groups. Males, who have higher infection rates, have larger activity spaces than females, and the four individuals who became infected during the study had significantly larger activity spaces than other participants. GPS tracking thus allowed me to identify activity space as a novel risk factor for leptospiral infection.

Another critical and as yet unanswered question about leptospirosis is whether there is naturally-acquired immunity to reinfection. Measuring this requires accurate infection histories, which we attempt to capture through longitudinal seroincidence studies. However, conventional methods of interpreting paired serology do not take into account the antibody titer decay between samples. Failing to do so sets an artificially high baseline titer from which to calculate a four-fold rise (the standard infection criterion) and reduces the probability an infection event will be recorded as such. Accounting for titer decay in leptospirosis is further complicated by the fact that re-exposure is a common feature of longitudinal data and that the gold standard serologic assay is interval-censored. I developed a flexible likelihood-based method to estimate titer decay using a dataset free of re-exposure. I then applied the decay rate to longitudinal cohort data from urban slum residents. When accounting for titer decay over a six-month interval, infections defined by a four-fold rise increased by a mean of 88%, and over a 12-month period the increase was 763%. Conventional serological interpretations thus severely underestimate leptospirosis infection, with the level of underestimation dependent on the sampling interval. The higher number of cases identified by our modified definition will provide the necessary data to evaluate evidence of acquired immunity due to a pre-existing infection.

This dissertation contributes to the understanding of leptospirosis transmission at both the regional and individual scale. It also provides information that can be used to target public health interventions and inform vaccine development. The tools developed herein, while used to answer questions about leptospirosis, have broader applications to a range of questions and pathogens.